24 TOWERS AND TANKS fOR WATER-WORKS. 



or less closely against the stand-pipe plates and did not fall pre- 

 vious to the initial rupture near the base. 



Among the countless fragments of ice were a number of very 

 large sheets or chunks, two of which fell upon the bed-plate 

 and another immediately in front of the gate-chamber door. 

 These masses and other boulder-like pieces of almost spherical 

 form had unquestionably floated on the top surface of the water 

 inside the ice tube, forming a broken sheet 30 inches to 3 feet 

 in thickness. With these masses floating at or near the top of 

 the stand-pipe, with the water surface held about stationary for 

 a short time during the previous night, a very slight formation 

 of ice, even less than that found on the plates, would weld them 

 into a self-supporting sheet. The failure of this ice roof with 

 the falling water-level and consequent atmospheic pressure from 

 above, accompanied by the morning rise of temperature, would 

 account for the crashing sound within the stand-pipe heard 

 the instant before the initial rupture occurred. The capacity 

 of the stand-pipe free from ice being about 500,000 gals, the 

 estimated volume of ice, assuming a shell 95 feet high, with 18- 

 inch average walls and a 3o-in. top sheet, was 14,190 cubic feet ; 

 which would reduce the capacity of the full tank to about 400,000 

 gals. With the water-level, say, at 72 feet above the base, the 

 volume of water in the pipe when the failure occurred was not 

 far from 300,000 gals, from 6 A.M. to 8 A.M., indicating a con- 

 sumption of perhaps 100,000 gals, or less during the two hours. 

 The total weight of ice in the stand-pipe, assuming these condi- 

 tions, was about 400 tons, of which 35 tons was in the top sheet, 

 whose fall is supposed to have preceded the failure. 



An examination of the material of which the pipe was com- 

 posed showed considerable irregularity, many fractures showing 

 more or less dead and laminated appearance and evidence of 

 brittleness, such as cracks and crystalline spots in the fractims. 

 Rivet fractures generally exhibited satisfactory material, although 

 the laying out showed poorly matched holes; there were also signs 



